I. Deborah Raji

I. Deborah Raji is an AI researcher, algorithmic auditor, and advocate whose work has focused on the practical assessment of AI system performance, bias, and accountability. She is a researcher at the University of California, Berkeley, and a fellow at the Mozilla Foundation, where she works on responsible AI policy and the development of technical standards for AI auditing.

Raji's research trajectory has been shaped by a practical question: not whether AI systems are theoretically biased, but whether the methods exist to detect that bias rigorously in deployed systems. Her early work on facial recognition, conducted while she was a student researcher in partnership with Joy Buolamwini at the MIT Media Lab, produced some of the most consequential empirical findings in AI accountability. Their research demonstrated that commercial facial recognition systems from major vendors, including IBM, Microsoft, and Face++, performed significantly worse on darker-skinned faces and on women, with error rates for darker-skinned women more than thirty percentage points higher than for lighter-skinned men.

That work directly influenced decisions by IBM, Microsoft, and Amazon to pause or restrict the sale of facial recognition software to law enforcement. It also influenced US legislative efforts to regulate biometric AI and contributed to a wider policy conversation about the deployment of AI systems in high-stakes contexts.

Raji's subsequent research has focused on developing the methodological infrastructure for AI accountability: how should organizations audit AI systems for bias and harm? What does an internal accountability framework look like? Her "Closing the AI Accountability Gap" paper, co-authored with researchers at Google and elsewhere, established a framework for end-to-end internal algorithmic auditing that has been widely cited.

She has also been a co-author on the foundational Model Cards paper that introduced standardized documentation practices for machine learning models, and on numerous subsequent works that have helped define what credible AI evaluation looks like in research and in industry. Outside academic publishing, Raji has testified before legislators, served on advisory bodies considering AI regulation, and written for general audiences on the public-interest stakes of AI accountability. Her work spans the rare combination of technical contribution to peer-reviewed research, policy translation for regulators, and movement-building inside the AI ethics community, and that range is part of why her endorsement of practitioner-oriented frameworks like the AI First Principles carries weight.

I. Deborah Raji's research is cited twice in the treatise, in two different principles, because her work addresses both the diagnosis and the remedy. Her "Closing the AI Accountability Gap" paper is cited in AI Inherits Messiness for introducing the concept of "bias bounties," organizations actively seeking to discover and document failure modes before they cause harm, as the solution framework for managing AI's inherited biases.

The same paper is cited again in AI Fails Silently for the concept of "circuit breakers," automated mechanisms that halt AI operation when the system detects it is operating outside validated parameters. Raji's contribution to the principles is methodological: she has built the technical vocabulary and frameworks for the kind of systematic, proactive auditing the principles require.